Baseless Next Generation Rumors with no Technical Merits [pre E3 2019] *spawn*

good for an anchor point, not sure if this is where their final production solution will end up. But good to note.

 

Servers can move a lot of heat . So that isn't always true

At the start MS will need to have hardware set up for cloud customers. So before the first console is even made ms will have already made tens if not hundreds of thousands of blade servers.

They would I assume move the hardware to cater to those not on 4k or might not need a premium experience and the refresh hardware would be for those who want the best.

 

This coming gen will be much more rounded design than last gen in everyway, that’s what we are looking at here,

 

Should we get a poll for Tflops prediction? Would be good to know what everyone's expectations are.

 

The thing is who gave AMD more money to design the custom APU?

The 2/3 engineer thing could be bogus or it could be that Sony put a lot more money AMDs way and it kind of makes sense because the PlayStation brand is very important to Sony and it generates a lot more cash than the Xbox division does for Microsoft. So if you look at it like that it makes sense for Sony to have put more resources into the custom chips that AMD are making for them and how the 2/3 engineer thing is probable.

 

Applicable as discusses zen 3 and other things that I think easily could be used in console in a baseless rumour thread anyway

 

Or the cloud and Azure is very important to MS (as well as gaming now) , so they invested a lot with AMD to produce a chip for cloud and console use.
It's also possible that R&D could be costed inside MS as 60 40 or something between Azure and Xbox department.
MS likes to move accounting/reporting around and this would help fund Xbox without looking bad on the balance sheet.
Also Xbox would be "selling" compute & processing to Azure as an ongoing income.

If they see gaming as more than just games and consoles but Azure and cloud it becomes a lot more important to them as a pillar.
Which then ends up being higher investment into gaming

 

Yes but AMD has already done a lot of research and got products for cloud because there's many more customers there. I think that they might of made some changes to the custom APU for the next Xbox to be more in line with there cloud offering not the other way round.

Do we even know if Microsoft are using Navi?

 

I doubt relative performance will have much to do with budgets for customization. Besides AMD being the platform of choice, I doubt we see much of a difference in hardware performance of either machine.

Power is going come down to:

1) what each platform wants to sell their console for 399, 499 or 599 and what AMD can provide at that price point (fwiw both companies are going to be reluctant to launch north of 499)

2) how much (if anything) they'd be willing to subsidize the MSRP (IMO not much)

I'm not sure what percentage of budget either company has for the next generation goes towards hardware customization anyway.

Sony will likely do some optimization for their VR application but they are going to have some cost to catch up with bc and MS may do something do optimize frame rate, or buying developers to round out the exclusives or invest in more cloud infrastructure.

Point being both companies are well positioned on some fronts and behind on others and it arguably makes more sense to address the trouble areas than simply try and invest in hardware customization.

You have to ask yourself, sitting in a board room making an argument about the hardware design, why would either company want to launch with significant losses just to win a PR war about power when history shows power is not what drives sales? And sitting in that same board room making that argument while advocating for streaming hardware at launch to compliment your high powered lead design... Frankly I'm skeptical either company sells hardware at a loss as a streaming box with better margins, lower cost of entry should address that segment of the market.

 

If it made sense for games.
They could've also included the AI processor from Hololens 2 for all we know, as it may make sense in cloud for AI/ML, but could be used for games, which may not of been included in console otherwise.

We don't know anything hardware related.
Could be gerbil powered.
We don't know cpu, gpu, memory, storage,
We're just making educated guesses as baseline and go crazy from there.
Example, expect it to be zen based. Reasonable educated guess but is it zen 2, 3, amount of cores, threads?

 

So it could be more based on Radeon instinct because of what they will use in Azure ?

Also I was just trying to say that the 2/3 engineer thing is a possibility if Sony ponied up the cash for it and it wasn't some preference or my little pony theory that previous poster were jesting about or intimating.

 

I would tend to exclude the gerbils option because of animal rights activists

 

Squirrels then?

 

It would be a pr nightmare. One of my sources suggested prana for ps5, leveraging the experience on soul powered cells.

 

The PR was similarly bad. Cell scaled from embedded devices right up to supercomputers, but the actual chips only shared a common instruction set and base architecture. The actual chips themselves were very different.

I don't know what server designs folks think console-class chips are going into with minuscule amounts of L1 and L2 cache but what I'm reading there seems to be a belief that from that one fab line, great chips are going into severs and the less capable chips go into consoles.

 

AWS uses off the shelf CPU, and actually has a line based on epyc, that in the next iteration will be a binned batch of the same chiplets that will be sold at 99$ in cheap pc.

 

For a 2016-2017 cash-starved AMD it shouldn't be surprising for Lisa Su to favor being on Sony's good graces than to better compete with nvidia in the PC discrete GPU market. The former's success depends only on the partners working towards a common goal and had cash up front from Sony, whereas the later would depend on a competitor's amount of success.

That said, putting RTG man-months on console semi-custom SoCs is a much safer investment than putting them on discrete GPUs.

Like everything, the preference would obviously come down to money. Safer investment vs. riskier investment.

 

Wouldn't it be more likely that the chips are primarily there for game streaming, but can also be used for enterprise clients during times of low (gaming) use? Once you have them sitting there you might as well use them if they can run useful workloads and they're power efficient enough.

I'd image they'd be used primarily for GPU compute, like Radeon Instinct GPUs. Similar performance to Mi60 and same ram size*, with fast SSD directly connected via PCI-E. I could see a few hundred thousand of those having a use.

*Edit: forgot that Mi60 is 32GB. Don't see a console matching that. Still, could probably get a lot done with, say, 16GB like Mi50.

 

I think it's a different type of server for different workloads to the traditional. As others say, there's game streaming, but besides that you could have perhaps an application on a dedicated machine rather than one server running many users.

It's more a question of "what server applications could a console design be used for?" rather than "how does this design compare to other server architectures?" Because with the latter, servers being huge CPUs with huge cache and crazy amounts of RAM, there's not a lot in common with consoles!

 

I think Sony can make a low power (low clocked) special IDLE state where you mine criptocurrency...